MLS-C01 : AWS Certified Machine Learning – Specialty : Part 05

MLS-C01 : AWS Certified Machine Learning – Specialty : Part 05

1. A Data Scientist is training a multilayer perception (MLP) on a dataset with multiple classes. The target class of interest is unique compared to the other classes within the dataset, but it does not achieve and acceptable recall metric. The Data Scientist has already tried varying the number and size of the MLP’s hidden layers, which has not significantly improved the results. A solution to improve recall must be implemented as quickly as possible.

   Which techniques should be used to meet these requirements?

   • Gather more data using Amazon Mechanical Turk and then retrain
   • Train an anomaly detection model instead of an MLP
   • Train an XGBoost model instead of an MLP
   • Add class weights to the MLP’s loss function and then retrain

2. A Machine Learning Specialist works for a credit card processing company and needs to predict which transactions may be fraudulent in near-real time. Specifically, the Specialist must train a model that returns the probability that a given transaction may fraudulent.

   How should the Specialist frame this business problem?

   • Streaming classification
   • Binary classification
   • Multi-category classification
   • Regression classification

3. A real estate company wants to create a machine learning model for predicting housing prices based on a historical dataset. The dataset contains 32 features.

   Which model will meet the business requirement?

   • Logistic regression
   • Linear regression
   • K-means
   • Principal component analysis (PCA)

4. A Machine Learning Specialist is applying a linear least squares regression model to a dataset with 1,000 records and 50 features. Prior to training, the ML Specialist notices that two features are perfectly linearly dependent.

   Why could this be an issue for the linear least squares regression model?

   • It could cause the backpropagation algorithm to fail during training
   • It could create a singular matrix during optimization, which fails to define a unique solution
   • It could modify the loss function during optimization, causing it to fail during training
   • It could introduce non-linear dependencies within the data, which could invalidate the linear assumptions of the model

5. Given the following confusion matrix for a movie classification model, what is the true class frequency for Romance and the predicted class frequency for Adventure?

   

   • The true class frequency for Romance is 77.56% and the predicted class frequency for Adventure is 20.85%
   • The true class frequency for Romance is 57.92% and the predicted class frequency for Adventure is 13.12%
   • The true class frequency for Romance is 0.78 and the predicted class frequency for Adventure is (0.47-0.32)
   • The true class frequency for Romance is 77.56% * 0.78 and the predicted class frequency for Adventure is 20.85%*0.32

6. A Machine Learning Specialist wants to bring a custom algorithm to Amazon SageMaker. The Specialist implements the algorithm in a Docker container supported by Amazon SageMaker.

   How should the Specialist package the Docker container so that Amazon SageMaker can launch the training correctly?

   • Modify the bash_profile file in the container and add a bash command to start the training program
   • Use CMD config in the Dockerfile to add the training program as a CMD of the image
   • Configure the training program as an ENTRYPOINT named train
   • Copy the training program to directory /opt/ml/train

7. A Data Scientist needs to analyze employment data. The dataset contains approximately 10 million observations on people across 10 different features. During the preliminary analysis, the Data Scientist notices that income and age distributions are not normal. While income levels shows a right skew as expected, with fewer individuals having a higher income, the age distribution also show a right skew, with fewer older individuals participating in the workforce.

   Which feature transformations can the Data Scientist apply to fix the incorrectly skewed data? (Choose two.)

   • Cross-validation
   • Numerical value binning
   • High-degree polynomial transformation
   • Logarithmic transformation
   • One hot encoding

8. A web-based company wants to improve its conversion rate on its landing page. Using a large historical dataset of customer visits, the company has repeatedly trained a multi-class deep learning network algorithm on Amazon SageMaker. However, there is an overfitting problem: training data shows 90% accuracy in predictions, while test data shows 70% accuracy only.

   The company needs to boost the generalization of its model before deploying it into production to maximize conversions of visits to purchases.

   Which action is recommended to provide the HIGHEST accuracy model for the company’s test and validation data?

   • Increase the randomization of training data in the mini-batches used in training
   • Allocate a higher proportion of the overall data to the training dataset
   • Apply L1 or L2 regularization and dropouts to the training
   • Reduce the number of layers and units (or neurons) from the deep learning network

9. A Machine Learning Specialist is given a structured dataset on the shopping habits of a company’s customer base. The dataset contains thousands of columns of data and hundreds of numerical columns for each customer. The Specialist wants to identify whether there are natural groupings for these columns across all customers and visualize the results as quickly as possible.

   What approach should the Specialist take to accomplish these tasks?

   • Embed the numerical features using the t-distributed stochastic neighbor embedding (t-SNE) algorithm and create a scatter plot.
   • Run k-means using the Euclidean distance measure for different values of k and create an elbow plot.
   • Embed the numerical features using the t-distributed stochastic neighbor embedding (t-SNE) algorithm and create a line graph.
   • Run k-means using the Euclidean distance measure for different values of k and create box plots for each numerical column within each cluster.

10. A Machine Learning Specialist is planning to create a long-running Amazon EMR cluster. The EMR cluster will have 1 master node, 10 core nodes, and 20 task nodes. To save on costs, the Specialist will use Spot Instances in the EMR cluster.

    Which nodes should the Specialist launch on Spot Instances?

    • Master node
    • Any of the core nodes
    • Any of the task nodes
    • Both core and task nodes

11. A manufacturer of car engines collects data from cars as they are being driven. The data collected includes timestamp, engine temperature, rotations per minute (RPM), and other sensor readings. The company wants to predict when an engine is going to have a problem, so it can notify drivers in advance to get engine maintenance. The engine data is loaded into a data lake for training.

    Which is the MOST suitable predictive model that can be deployed into production?

    • Add labels over time to indicate which engine faults occur at what time in the future to turn this into a supervised learning problem. Use a recurrent neural network (RNN) to train the model to recognize when an engine might need maintenance for a certain fault.
    • This data requires an unsupervised learning algorithm. Use Amazon SageMaker k-means to cluster the data.
    • Add labels over time to indicate which engine faults occur at what time in the future to turn this into a supervised learning problem. Use a convolutional neural network (CNN) to train the model to recognize when an engine might need maintenance for a certain fault.
    • This data is already formulated as a time series. Use Amazon SageMaker seq2seq to model the time series.

12. A company wants to predict the sale prices of houses based on available historical sales data. The target variable in the company’s dataset is the sale price. The features include parameters such as the lot size, living area measurements, non-living area measurements, number of bedrooms, number of bathrooms, year built, and postal code. The company wants to use multi-variable linear regression to predict house sale prices.

    Which step should a machine learning specialist take to remove features that are irrelevant for the analysis and reduce the model’s complexity?

    • Plot a histogram of the features and compute their standard deviation. Remove features with high variance.
    • Plot a histogram of the features and compute their standard deviation. Remove features with low variance.
    • Build a heatmap showing the correlation of the dataset against itself. Remove features with low mutual correlation scores.
    • Run a correlation check of all features against the target variable. Remove features with low target variable correlation scores.

13. A company wants to classify user behavior as either fraudulent or normal. Based on internal research, a machine learning specialist will build a binary classifier based on two features: age of account, denoted by x, and transaction month, denoted by y. The class distributions are illustrated in the provided figure. The positive class is portrayed in red, while the negative class is portrayed in black.

    

    Which model would have the HIGHEST accuracy?

    • Linear support vector machine (SVM)
    • Decision tree
    • Support vector machine (SVM) with a radial basis function kernel
    • Single perceptron with a Tanh activation function

14. A health care company is planning to use neural networks to classify their X-ray images into normal and abnormal classes. The labeled data is divided into a training set of 1,000 images and a test set of 200 images. The initial training of a neural network model with 50 hidden layers yielded 99% accuracy on the training set, but only 55% accuracy on the test set.

    What changes should the Specialist consider to solve this issue? (Choose three.)

    • Choose a higher number of layers
    • Choose a lower number of layers
    • Choose a smaller learning rate
    • Enable dropout
    • Include all the images from the test set in the training set
    • Enable early stopping

15. This graph shows the training and validation loss against the epochs for a neural network.

    The network being trained is as follows:
    – Two dense layers, one output neuron
    – 100 neurons in each layer
    – 100 epochs
    – Random initialization of weights

    

    Which technique can be used to improve model performance in terms of accuracy in the validation set?

    • Early stopping
    • Random initialization of weights with appropriate seed
    • Increasing the number of epochs
    • Adding another layer with the 100 neurons

16. A Machine Learning Specialist is attempting to build a linear regression model.

    

    Given the displayed residual plot only, what is the MOST likely problem with the model?

    • Linear regression is inappropriate. The residuals do not have constant variance.
    • Linear regression is inappropriate. The underlying data has outliers.
    • Linear regression is appropriate. The residuals have a zero mean.
    • Linear regression is appropriate. The residuals have constant variance.

17. A large company has developed a BI application that generates reports and dashboards using data collected from various operational metrics. The company wants to provide executives with an enhanced experience so they can use natural language to get data from the reports. The company wants the executives to be able ask questions using written and spoken interfaces.

    Which combination of services can be used to build this conversational interface? (Choose three.)

    • Alexa for Business
    • Amazon Connect
    • Amazon Lex
    • Amazon Polly
    • Amazon Comprehend
    • Amazon Transcribe

18. A machine learning specialist works for a fruit processing company and needs to build a system that categorizes apples into three types. The specialist has collected a dataset that contains 150 images for each type of apple and applied transfer learning on a neural network that was pretrained on ImageNet with this dataset.

    The company requires at least 85% accuracy to make use of the model.

    After an exhaustive grid search, the optimal hyperparameters produced the following:
    – 68% accuracy on the training set
    – 67% accuracy on the validation set

    What can the machine learning specialist do to improve the system’s accuracy?

    • Upload the model to an Amazon SageMaker notebook instance and use the Amazon SageMaker HPO feature to optimize the model’s hyperparameters.
    • Add more data to the training set and retrain the model using transfer learning to reduce the bias.
    • Use a neural network model with more layers that are pretrained on ImageNet and apply transfer learning to increase the variance.
    • Train a new model using the current neural network architecture.

19. A company uses camera images of the tops of items displayed on store shelves to determine which items were removed and which ones still remain. After several hours of data labeling, the company has a total of 1,000 hand-labeled images covering 10 distinct items. The training results were poor.

    Which machine learning approach fulfills the company’s long-term needs?

    • Convert the images to grayscale and retrain the model
    • Reduce the number of distinct items from 10 to 2, build the model, and iterate
    • Attach different colored labels to each item, take the images again, and build the model
    • Augment training data for each item using image variants like inversions and translations, build the model, and iterate.

20. A Data Scientist is developing a binary classifier to predict whether a patient has a particular disease on a series of test results. The Data Scientist has data on 400 patients randomly selected from the population. The disease is seen in 3% of the population.

    Which cross-validation strategy should the Data Scientist adopt?

    • A k-fold cross-validation strategy with k=5
    • A stratified k-fold cross-validation strategy with k=5
    • A k-fold cross-validation strategy with k=5 and 3 repeats
    • An 80/20 stratified split between training and validation